Skip to main content

Full text of "Why not a political Coase Theorem? : social conflict, commitment and politics"

See other formats


MIT LIBRARIES DUPL 





* / \ * 

LIBRARIES 



Digitized by the Internet Archive 

in 2011 with funding from 

Boston Library Consortium Member Libraries 



http://www.archive.org/details/whynotpoliticalcOOacem 



w 







Massachusetts Institute of Technology 

Department of Economics 

Working Paper Series 



Why Not a Political Coase Theorem? 
Social Conflict, Commitment and Politics* 



Daron Acemoglu 



Working Paper 02-44 
November 19,2002 



MASSACHUSETTS INSTITUTE™! 
OF TECHNOLOGY 

1 




LIBRARIES 



RoomE52-251 

50 Memorial Drive 

Cambridge, MA 02 142 



This paper can be downloaded without charge from the 
Social Science Research Network Paper Collection at 
http://papers.ssrn.com/abstract=355900 



Why Not A Political Coase Theorem? Social 
Conflict, Comrnitment and Politics* 

Daron Acemoglu 

Department of Economics 

Massachusetts Institute of Technology 

50 Memorial Drive 

Cambridge, MA 02142 

November 19, 2002 



Abstract 

Do societies choose inefficient policies and institutions, in contrast to what would 
be suggested by a reasoning extending the Coase Theorem to politics? Do societies 
choose inefficient policies and institutions because of differences in the beliefs and 
ideologies of their peoples or leaders? Or are inefficiencies in politics and economics 
the outcome of social and distributional conflicts? This paper discusses these var- 
ious approaches to political economy, and develops the argument that there are 
strong empirical and theoretical grounds for believing that inefficient policies and 
institutions are prevalent, and that they are chosen because they serve the interests 
of politicians or social groups holding political power, at the expense of the society 
at large. At the center of the theoretical case are the commitment problems inher- 
ent in politics: parties holding political power cannot make commitments to bind 
their future actions because there is no outside agency with the coercive capacity 
to enforce such arrangements. 

Keywords: Political Economy, Institutions, Commitment, Social Conflict, Be- 
lief Differences, Appropriate Institutions, Economic Development, Colonialism. 

JEL Numbers: H2, N10, N40, 01. 



*I thank Raphael Auer for outstanding research assistance and seminar participants at the World 
Bank Appropriate Institutions conference and at the International Society for New Institutional Eco- 
nomics conference, and Simeon Djankov, Simon Johnson, James Robinson and two anonymous referees 
for comments. 



1 Introduction 

There is increasing interest in the effects of government policies, economic, political and 
legal institutions, and more broadly, of the organization of society on economic outcomes. 
There is also a sense among economists and social scientists that differences in policies 
and institutions are a first-order determinant of the very different economic and social 
fortunes of various countries, though we are far from a consensus on this point. Despite 
the range of important contributions on these topics (see, for example, the surveys in 
Austen-Smith and Banks, 1999, and Persson and Tabellini, 2000), we still do not know 
how to organize our thinking on the determinants of policies and institutions. 1 Why do 
certain societies choose different policies, different institutions, radically different ways of 
organizing their lives? This paper is an attempt to provide a simple taxonomy, and argue 
for the relevance of a particular approach. 

The taxonomy consists of grouping various approaches into three categories: 

1. Political Coase Theorem (PCT): the Coase Theorem maintains that when property 
rights are well-defined and there are no "transaction costs", economic agents will 
"contract" to achieve an efficient (output- or surplus-maximizing) outcome, irre- 
spective of who has the property rights on particular assets (Coase, 1960, Stigler, 
1966). An extension of this reasoning to the political sphere would suggest that 
political and economic transactions create a strong tendency towards policies and 
institutions that achieve the best outcomes given the varying needs and require- 
ments of societies, irrespective of who, or which social group, has political power. 
According to this approach, policy and institutional differences are not the major 
determinant of the differences in economic outcomes, since societies choose, at least 
approximately, the "right" policies and institutions given their conditions. 

2. Theories of Belief Differences (Modified Political Coase Theorem) societies may 



There is an important distinction between policies and institutions. We can loosely think of policies 
as choices made within a given political and social structure, e.g., the tax rate and fiscal policies. In 
contrast, institutions can be thought as determinants of the political and social structure that are more 
durable, and as such, constrain future choices and policies, e.g., whether the society is democratic or not, 
the nature of the legal constraints on the government, or the extent of private property rights enforcement. 
Although institutions are often predetermined at the point in time when certain policy choices are made, 
they are also chosen by the society. For example, governments and citizens decide what legal code will 
apply, and how stringently it will be enforced. The focus here is on why both inefficient policies and 
inefficient institutions are chosen and continue to be chosen. So for most of this paper I will not make 
much of a distinction between policies and institutions. 



choose different policies, with very different implications, because they — or their 
leaders — disagree about what would be good for the society. According to this 
approach, there is sufficient uncertainty about the right policies and institutions 
that well-meaning political actors differ about what is good for their own people. 
Societies where the leaders or the electorate turn out to be right ex post are those 
that prosper. The important point is that, just as with the PCT, there are strong 
forces preventing the implementation of policies that are known to be bad for the 
society at large, hence the label Modified Political Coase Theorem. 

3. Theories of Social Conflict: societies choose different policies, some very disastrous 
for their citizens, because those decisions are made by politicians or politically pow- 
erful social groups that are interested in maximizing their own payoffs, not aggregate 
output or social welfare. This category includes both theories where internal conflict 
within the society leads to inefficient choices and those where inefficient institutions 
and policies are imposed on societies from the outside (e.g., by colonial powers). 2 

At some level, the major divide is between Theories of Social Conflict, which em- 
phasize the prevalence of systematically inefficient government policies and institutional 
arrangements, versus the first two approaches, which stress the presence of social forces 
that rule out these types of inefficiencies. The rest of the paper argues that the PCT, 
in its simple form, or in its modified version built on belief differences, is not an appro- 
priate framework for thinking about policy and institutional differences across countries. 

2 This taxonomy implicitly classifies a lot of interesting theories that combine features from the three 
groups, while still giving a prominent role to social conflict, in the category of Theories of Social Conflict. 
Most important, in many theories featuring social conflict, societies will make different choices because 
of differences in their economic conditions, but generally there will not be strong forces towards efficient 
outcomes in any of these societies. 

A number of interesting interactions are also worth mentioning briefly. For example, certain groups 
may attempt, or manage, to convince others that their most-preferred policies also benefit the society at 
large (see Coate and Morris, 1995, for a model with this flavor). Another interesting interaction arises 
when some societies choose different policies initially because of differences in beliefs, but then these 
policies create or strengthen their own political constituencies, supporting the continued implementation 
of these inefficient policies (see Acemoglu, Aghion and Zilibotti, 2002, for a model with this flavor). 

Another set of approaches, especially popular among sociologists and political scientists, is also worth 
mentioning here. These also maintain that many policies and institutions are inefficient, but contrary to 
Theories of Social Conflict, they do not attempt to explain these inefficiencies by the economic or social 
objectives of competing groups. Instead, institutions and policies are presumed to arise as unintended 
consequences of other interactions. A salient example is Tilly's (1990) work, and its extension by Herbst 
(2000), which stresses the importance of the emergence of the nation-state for economic development, 
but whether the nation-state emerges or not is a consequence of other unrelated factors, for example 
population density or the frequency of wars. 



Existing evidence suggests that societies often choose inefficient policies and institutions, 
and in most cases they do this not because of differences in beliefs, but because of severe 
misalignments in the economic interests of politically decisive actors and the rest of the 
society. So Theories of Social Conflict appear to provide the right starting point for an 
analysis of policy and institutional differences. 

A major challenge for Theories of Social Conflict, however, is to pinpoint what spe- 
cific "transaction costs" would systematically prevent the Political Coase Theorem from 
applying. In other words, why do politicians and powerful social groups not make a deal 
with the rest of the society to choose the policies and institutions that maximize output 
(or social welfare) , and then redistribute part of the gains to themselves? Put even more 
strongly; why do powerful groups not "predate" efficiently? The theoretical analysis is 
intended to highlight some of the issues that arise in thinking about these questions. 

The argument I develop in Sections 4 and 5 of the paper is that, although the PCT may 
be a useful benchmark, its applicability is limited because of the inherent commitment 
problems associated with political power. Underlying the Coase Theorem is the ability to 
write enforceable contracts. Any enforcement problem will therefore potentiaDy limit the 
applicability of the Coase Theorem. 3 In the context of the PCT, there is a natural reason 
for widespread enforcement problems. Most contracts are enforced by "the state" . When 
it comes to contracts that the state or social groups controlling the state would like to 
write with others (e.g., the citizens), they will, by definition, be non-enforceable because 
groups controlling the state cannot commit not to using their power to renege on their 
promises or not to changing the terms of the contract. This implies that the allocation 
of political power creates an inherent commitment problem, undermining the potential to 
reach efficient outcomes. 

The commitment problem associated with the PCT is twofold: first, those in power, 

e.g., the rulers, cannot commit to not using this power — as long as they do not relinquish 

it — in ways that benefit them in the future. Second, if the rulers relinquish their power, 

the citizens cannot commit to making side payments to them in the future, because the 

3 Enforcement problems may arise from incomplete information, contracting costs or bounded ratio- 
nality (e.g., Anderlini and Felli, 1998, Dixit and Olson, 2000, Farrell, 1987, McKelvey and Page, 1999). I 
focus on the commitment problems here, because I believe these are much more central when it comes to 
the PCT. The recent literature on transaction costs and the organization of the firm (e.g., Williamson, 
1981, Grossman and Hart, 1986, Hart and Moore, 1990) similarly focuses on why the distribution of 
property rights may matter for incentives when contracts are incomplete. There are a number of informal 
attempts to extend the reasoning of transaction costs to politics, see, for example, North (1990) or Spiller 
and Tommasi (2002). 



former rulers no longer possess the political power to enforce such promises. This double 
commitment problem restricts the potential remedies to inefficiencies. Nevertheless, be- 
cause the relationship between the state and the citizens is repeated, there may be some 
amount of commitment based on reputation — supported by the threat of future punish- 
ments. As a result, the extent to which the PCT will provide a good approximation to 
reality depends on the possibility of commitment via constitutions or other institutions 
and on how good a substitute this type of reputation-based commitment is for enforceable 
contracts, as well as on the extent of distributional conflict between various social groups 
in society. 

To focus on the inherent commitment problems present in political situations, the sim- 
ple model analyzed here allows unrestricted transfers and taxes, including non-distortionary 
lump-sum taxes. The inefficiencies arise not because of some restrictions on the technology 
of taxation, but because of the political-economic interactions between different groups 
and agents. Interestingly, despite the availability of non-distortionary taxes, the model 
also suggests that in this type of repeated game environment, the equilibrium may involve 
distortionary taxes. The reason is that the allocation has to satisfy the incentive com- 
patibility constraint of the ruler, which requires the current output not to be too large; 
otherwise, the ruler would prefer to grab all the output rather than stick to the agreement. 
With lump-sum taxes, individuals are the residual claimants of the returns they generate 
from their investments, and will have a tendency to "overinvest" , violating the incentive 
compatibility constraint of the ruler. Distortionary taxation may then be necessary to 
guarantee levels of investment consistent with the incentive compatibility constraint of 
the ruler. 

There is a large literature on distortionary policies of governments, which is nicely 
surveyed in Robinson (1998), with a similar distinction to the one here between bad 
policies that arise due to belief differences and those originating from social conflict. The 
most celebrated models of distortionary policies are the voting models, where the median 
or the decisive voter may choose policies that redistribute resources from the society 
as a whole to himself or to his group (e.g., Romer, 1975, Meltzer and Richards, 1981, 
or see Persson and Tabellini, 2000). There is also a large literature in political science 
on how voting behavior and the organization of parties interact to produce equilibrium 
policies (see, for example, Aldrich, 1983, Baron and Ferejohn, 1989, Dixit and Londregan, 
1995, Myerson, 1995, Snyder, 1990). A range of other papers emphasize conflict between 



bureaucrats or politicians and the society, for example, Buchanan and TuUock (1962), 
Ferejohn, (1986), Persson, Roland and Tabellini (1997), and Shleifer and Vishny (1998). 
These papers do not consider why politically powerful groups cannot extract resources 
from the rest of the society in an efficient manner, however. In fact, much of this literature 
rules out efficient methods of redistribution and takes it for granted that rent-maximizing 
behavior by rulers or the government will result in inefficiencies. The focus here is instead 
on why efficient policies fail to arise. In this respect, this paper is related to North 
(1981), Libecap (1989) and Olson (2000), who emphasize inefficient policies resulting from 
distributional conflicts. North, for example, suggests that rulers will choose the system of 
property rights in order to maximize their return, and this will lead to inefficiencies, but 
he also places considerable emphasis on differences in beliefs. Neither North nor Libecap 
nor Olson is explicit, however, on why a version of the Political Coase Theorem would 
not apply. 

By providing a rationale for inefficient methods of taxation, this paper also relates to a 
few existing studies investigating the reasons why, in many instances, societies use ineffi- 
cient redistributive policies rather than lump-sum taxation and transfers. Rodrik (1986), 
Wilson (1991) and Becker and Mulligan (1998) argue that if the amount of redistribution 
is endogenous, then politicians might want to commit to using inefficient methods in order 
to reduce total redistribution. Coate and Morris (1995) argue that inefficient redistribu- 
tion arises when politicians exploit the uncertainty of voters regarding which policies are 
efficient. In this context, most closely related are Besley and Coate (1998), Acemoglu 
and Robinson (2000b and 2002), Rajan and Zingales (2000), and Acemoglu and Robinson 
(2001). Besley and Coate analyze a two-period political economy model, and show how 
certain types of "inefficiencies" may arise because efficient policies would affect the iden- 
tity of who is in power (though they limit the politicians to linear taxes), and similarly 
emphasize the importance of commitment problems. 4 Acemoglu and Robinson (2000b 
and 2002) develop a theory where elites may want to block the introduction of new and 
efficient technologies because this will reduce their future political power, and similarly, 
Rajan and Zingales (2000) show how organizations make inefficient choices because each 

4 Other related papers emphasizing the importance of commitment issues in politics include: North and 
Weingast (1989), who argue that the introduction of the English Parliament in the seventeenth century 
was a commitment to low taxes in the future; Weingast (1998) who interprets the Missouri compromise 
as a commitment by Northerners not to attempt to abolish slavery in the South: and finally Acemoglu 
and Robinson (2000a), who argue that the introduction of democracy was a commitment by the rich elite 
to future redistribution. 



group (or agent) is worried that others in the organization will get richer, and demand 
more concessions in the future. Finally, Acemoglu and Robinson show that inefficient 
methods of redistribution, rather than more efficient alternatives, may arise as a method 
of maintaining future political power (see also Persson and Svensson, 1989, and Aghion 
and Bolton, 1990, on the use of fiscal policy to affect future elections). None of these pa- 
pers address the more general issue of how commitment problems undermine the "Political 
Coase Theorem" , which is the focus of the current paper, nor do they analyze repeated 
games where punishment strategies may substitute for lack of formal commitment. 5 

The rest of the paper is organized as follows. The next section revisits the above tax- 
onomy of various approaches to the determination of pohcies and institutions introduced 
above. Section 3 argues that in practice, neither the PCT nor the Modified PCT provide a 
satisfactory framework for studying cross-country differences in institutions and policies. 
Sections 4 and 5 analyze a simple model of conflict between the ruler and the citizens, 
highlight the commitment problems inherent in political transactions, and show why the 
reasoning of the PCT will not apply in general. This analysis also develops some simple 
comparative statics from the model, and shows why distortionary taxes may be neces- 
sary to reduce overinvestment by citizens, which would otherwise violate the incentive 
compatibility constraint of rulers. 

2 A Simple Taxonomy 

To emphasize the differences between various approaches and build a simple taxonomy, 
consider the following setup, with Y denoting aggregate output or consumption, which I 
take to represent social welfare (thus avoiding some of the complications that come from 
Pareto comparisons, and focusing on the main point here). Moreover, suppose that we 
can write 

Y = F(X,P), 

where X is a vector of economic, geographic, social or other characteristics that are taken 

as given and directly influence economic outcomes, and P is a vector of pohcies and 

institutions that can potentially affect the outcomes of interest. I define P(. | X) as the 

5 In focusing on infinite-horizon models with self-enforcing arrangements, this model is also similar to 
Dixit, Grossman, and Gul (2000) who analyze self-enforcing political deals between groups with different 
interests. 



set of policies that maximize output, given a vector of characteristics X, i.e., 
P*{X)£P(.\X) <*=> P*{X) e argmaxF(X,P). 

The Political Coase Theorem, maintains that there are strong forces leading societies 
towards some P* {X) in P(. | X). The underlying idea is that if a society is pursuing 
a policy P (X) g P(. | X), then a switch to P* {X) G P(. | X) will create aggregate 
gains. If these gains correspond to a Pareto improvement, then all political systems will 
implement this change. If the change creates only a potential Pareto improvement, then 
part of the gains can be redistributed to those that, are losing out via various mechanisms, 
or at the very least, the winners can lobby or vote for the beneficial change. A number 
of social scientists have proposed limited forms of this Political Coase Theorem. Becker 
(1983, 1985), for example, pointed out how competition between pressure groups could 
create a force towards efficient policies. Wittman (1995) pushed this argument further and 
formulated an informal Political Coase Theorem for democratic societies. Wittman argued 
that with rational voters, democratic societies generally produce Pareto efficient, even 
wealth-maximizing, outcomes. In fact, Wittman's argument relies little on democratic 
institutions, and his reasoning could even apply to nondemocratic societies. 

To the extent that P(. | X) is not a singleton, we can observe considerable policy 
differences across two identical societies, but the performance of these two societies should 
not be appreciably different. An example could be differences in policies regarding the 
role of the government in the economy between the Anglo-Saxon economies, in particular, 
the U.S. and the UK, and Continental European countries, which do not seem to lead to 
major differences in the economic performance between these two sets of countries. 6 

When we look at a cross-section of societies in the data, however, we also see more 

major differences in policies and institutions, for example, free-market policies in some 

societies like Hong Kong, and heavy government involvement and widespread corruption 

in some others like Indonesia. But according to the PCT, various government interventions 

and corruption in Indonesia are not the reason why this country is poorer than Hong Kong. 

Each is choosing the policies and institutions that are appropriate for their own situations, 

but they achieve different outcomes because their situations, their X's, are different. 

More specifically, for two societies with characteristics X and X' ^ X, we typically have 

F{X,P*(X)) ^ F(X',P*{X')), and moreover, F{X,P*{X)) > F(X,P*{X')) and 

See Hall and Soskice (2001) for a discussion of the costs and benefits of various different types of 
capitalism. 



F(X',P* {X')) > F(X',P* {X)). Thus, the PCT suggests that Indonesian institutions 
are not chosen inefficiently, but appropriately for their circumstances. 7 

This discussion implies that to refute the applicability of the Political Coase Theorem, 
we need to find systematic evidence that there are societies choosing P while F (X, P) < 
F (X, P') for some feasible alternative P', or simply that P ^ P (. | X). That is, we need 
to show that there are societies that persistently pursue wrong policies, with significant 
output and welfare consequences. 8 

Theories of Belief Differences (Modified PCT), on the other hand, emphasize that 
some subset of X, X u , is uncertain. To simplify the notation while elaborating on this, 
suppose that P(. | X) is a singleton, in particular P(. | X) = P* (X). Moreover, imagine 
that X = (X„ X u ), and suppose that P* (X c , X u ) ^ P* (X c , X' u ) whenever X u =£ X' u , that 
is, these uncertain characteristics affect which policies are right for the society. Suppose 
that politicians (or the society at large) have beliefs, denoted by G (X u ), over the actual 
distribution of X u . Also suppose that social welfare maximization corresponds to the 
maximization of expected aggregate output. Then define 



P* {X c , G) e arg max I F {X c , X u , P) dG. 



Now two societies with the same X c , and the same ex post realization of X u , may choose 
different policies because their ex ante beliefs over the payoff-relevant characteristics, the 
X u 's, are different. Given a particular realization of X u , some societies among those with 
the same X c and X u will be richer than others, i.e., typically F (X C ,X U ,P* (X C ,G)) ^ 
F(X Cl X u ,P*(X c ,G')){orG^G'. 

For example, the North Koreans may be choosing socialist policies and government 
ownership because they believe those are the policies that will increase welfare, while South 
Korea, which presumably had the same characteristics, X c and X u , chose a capitalist 
development path. Ex post, the South Koreans turned out to be right, hence they were 
the ones who adopted the right policies, and the ones who prospered, while North Koreans 
today suffer poverty and famine. 9 

7 See Glaesar and Shleifer (2002) for an explanation for why Britain and France chose very different 
legal codes and systems that were appropriate to their underlying circumstances. 

8 Throughout by "refuting" the PCT, I mean showing that there are significant and quantitatively 
important inefficiencies in the institutions and policies of some societies — of course, this definition poses 
the question of what is "significant and quantitatively important" . A refutation of the PCT does not 
imply that there are no forces towards more efficient arrangements. 

9 An interesting theory of policy differences arising from belief differences is developed by Piketty 



To refute the class of models in this group, we need to show that there are societies 
that pursue policies that could not be the right policies under any plausible scenario. In 
other words, denoting the set of admissible beliefs by G, if, for two feasible policies, P 
and P', fF{X c ,X u ,P')dG > J F (X c , X u , P) dG for all G G G, then we should never 
observe P. 

Finally, according to the Theories of Social Conflict, societies often, knowingly, choose 
some policy vector P(X) $• F(.\X), because policies and institutions are chosen to 
maximize the payoffs of those who hold political power, not to maximize social welfare or 
aggregate income. To emphasize the difference between this approach and the Political 
Coase Theorem, imagine another vector of variables Z, which do not directly affect Y, thus 
P* (X) is independent of Z. These variables may nonetheless influence the "equilibrium" 
policy, so we can have P (X, Z). Changes in Z will have no direct effect on output, but 
may have a powerful indirect impact by influencing the gap between P (X, Z) and P* (X). 
In other words, we need to find a variable, Z, that is like an instrument in econometrics: 
it influences X, but has no direct effect on F. 

At this level of generality, Theories of Social Conflict are more like a residual group; 
if we can show that certain societies systematically, and knowingly, pursue inefficient 
policies, we are in the realm of Theories of Social Conflict. But the usefulness of these 
theories depends, in turn, on whether they can pinpoint an interesting mechanism for why 
political and economic bargains are not struck to achieve better policies and institutions 
(i.e., what are the salient "transaction costs" preventing the PCT from applying?), and 
whether we can identify a range of institutional or other social variables, the Z's, that 
affect the degree of inefficiency of policies. 

3 What the Data Say 

In this section, I briefly develop the argument that cross-country differences in policies 

and institutions are important determinants of economic performance, and the origins 

of these differences do not he in different perceptions of the peoples and the leaders, but 

in the social conflicts that exist between these leaders, or the social groups that these 

leaders represent, and the rest of the society. As noted above, the purpose of this exercise 

(1995) where individuals vote over the degree of redistribution in the economy as a function of their 
beliefs on the importance of individual effort in economic success. These beliefs, in turn, evolve as a 
result of various economic interactions and tax policies. See Romer (1997) and Mukand and Rodrik 
(2002) for recent studies arguing for the importance of these issues. 



is not to argue that there are no economic and political forces towards more efficient social 
arrangements, but to show that there are salient examples of inefficient institutions and 
policies, accounting for quantitatively large variations in economic performance. 

Recall also that societies may choose inefficient policies and institutions both because 
of internal conflict and because these choices are imposed on them externally. Although 
inefficiencies arising from internal conflict are at least as important, in the latter part of 
this section I focus on two examples of inefficient institutions imposed by outside forces, 
because these episodes make it clear that these institutional choices were not in response 
to different economic circumstances (i.e., they exploit sources of exogenous variation from 
history). 

3.1 Differences in Institutions and Economic Outcomes 

There are tremendous cross-country differences in the way that economic and political 
life is organized. Let us focus here on a range of characteristics which we can think of 
as "economic institutions," for example, the degree of stable property rights enforcement, 
the extent of equal opportunity and of entry barriers. A voluminous literature documents 
large cross-country differences in economic institutions, and a strong correlation between 
these institutions and economic performance. 

To pick a few examples, Knack and Keefer (1995) look at measures of property rights 
enforcement compiled by international business organizations (in particular Political and 
Risk Services), Mauro (1995) looks at measures of corruption, and Djankov, La Porta, 
Lopez-De-Silanes and Shleifer (2002) compile measures of entry barriers across countries, 
while many studies look at variation in educational institutions and the corresponding 
differences in human capital (e.g., Ringer, 1979, Krueger and Lindahl, 2001, Hanushek 
and Kimko, 2000). All of these authors find substantial differences in these measures 
of economic institutions, and significant correlation between these measures and various 
indicators of economic performance. For example, Djankov et al. document that, while the 
total cost of opening a medium-size business in the United States is less than 0.02 percent 
of GDP per capita in 1999, the same cost is 2.7 percent of GDP per capita in Nigeria, 1.16 
percent in Kenya 0.91 percent in Ecuador and 4.95 percent in the Dominican Republic. 
These entry barriers are highly correlated with various economic outcomes, including the 
rate of economic growth and the level of development. 

A defender of the PCT could counter these empirical patterns with the following 



10 



argument: this type of correlation does not establish that countries are choosing the 
wrong institutions. After all, the United States differs from Nigeria, Kenya and the Do- 
minican Republic in its economic characteristics, i.e., its A's. And different A's require 
different optimal (appropriate) policies and institutions. In terms of the notation intro- 
duced above, it may be the case that X ^ X', and consequently, P* (X) ^ P* (A"'), and 
F (X, P* (X)) ^ F (X 1 , P* (A')), so we might simply be observing the optimal response 
of different societies to their own varying conditions. Perhaps, given the circumstances in 
the Dominican Republic, it is not worth investing in the arrangements to reduce the costs 
of opening and doing business (or in the modified form of the PCT, perhaps the people 
of the Dominican Republic believe that high entry barriers are good for the society). 10 

To refute the general applicability of the PCT for analyzing differences in institutions 
and policies across countries, and their impact on economic outcomes, we need to show 
that otherwise identical, or at least similar, societies choose different institutions and 
policies because of reasons that do not directly affect economic outcomes, and experience 
differential economic performances as a result of these choices. This is essentially the 
reasoning of instrumental variables. Therefore, to refute the PCT, we have to find a 
source of variation, the Z's, that do not directly influence economic outcomes, but affect 
the choice of policies and institutions, and then show that these differences matter for 
economic outcomes. In other words, we have to find some type of natural (or "unnatural" ) 
social experiment where for political or historical reasons some societies end up with very 
different institutions than others (in addition, if we also want to refute the modified PCT, 
we would have to show that the variation captured by the Z's is not working solely 
through belief differences). Naturally, the focus on historical and sources of variation 
in institutions and policies does not mean that these provide the most major reason for 
cross-country differences. Internal dynamics leading to different policies institutions are 
likely to be at least as important, but for the purposes of this exercise, external sources 
of variation make identification easier. 



10 An example of optimal non-enforcement of private property rights may be the case of North American 
Indians before the eighteenth century. Demsetz (1967) argues that despite the potential for overhunting 
of game, the costs of enforcing property rights in land were higher than the benefits, since without the fur 
market, there were only weak incentives for overhunting. This changed after the Indians started trading 
fur with the white Americans, at which point the incentives for overhunting and the costs of lack of 
property rights increased, and private property rights in land were duly introduced. 



II 



3.2 Colonialism and Institutional Development 

European colonization of the rest of the world provides the best laboratory (almost a 
natural experiment) to investigate these issues. From the late 15th century, Europeans 
dominated and colonized much of the rest of the Globe. Together with European domi- 
nance came the imposition of various types of institutions in the colonies. Most interesting 
for our purposes, Europeans imposed very different institutions and social power struc- 
tures in different parts of the world. 

Acemoglu, Johnson and Robinson (2001) document that in a large number of colonies, 
especially those in Africa, Central America, the Caribbean and South Asia, European 
powers set up "extractive states". These institutions did not introduce much protection 
for private property, nor did they provide checks and balances against government ex- 
propriation. The explicit aim of the Europeans in these colonies was the extraction of 
resources, in one form or another. In the Caribbean, this took the form of slave planta- 
tions, in parts of Central and Meso America, mining based on forced labor. In Africa, 
Europeans were first interested in the extraction of slaves to employ on the plantations in 
the Americas, and later developed other methods of extracting resources, including high 
taxes and extraction of natural resources. 11 Other economic institutions that Europeans 
set up in these colonies were similarly detrimental to economic advancement; there was 
little investment in the human capital of the majority of the population, and access to 
key resources was concentrated in the hands of a few. 

This colonization strategy and the associated institutions contrast with the institutions 
Europeans set up in colonies where they settled in large numbers, for example, the United 
States, Canada, Australia and New Zealand. In these colonies, life was modeled after the 
home country, and the emphasis was on the enforcement of property rights for a broad 
cross-section of the society, especially smallholders, merchants and entrepreneurs. See for 
example Gann and Duignan (1962), Robinson and Gallagher (1961), Denoon (1983), Cain 
and Hopkins (1993). 

n For example, Davis and Huttenback (1986, p. 307) calculate that before 1885, investment in the 
British empire had a return 25 percent higher than that on domestic investment. Roberts (1976, p. 
193) calculates a large transfer of resources from Northern Rhodesia to Britain, in return for minimal 
investment. Manning (1982) estimates that between 1905 and 1914, 50 percent of GDP in Dahomey was 
extracted by the French, Young (1994, p. 125) notes that taxation rates in Tunisia were four times higher 
than those in metropolitan France, and Peemans (1975) documents the amount of resources extracted 
from the Belgian Congo and calculates that tax rates on Africans approached 60 percent of their income 
during the 1920's and 1930's. 



12 



Acemoglu, Johnson and Robinson (2001) document that the crucial determinant of 
whether Europeans chose the path of extractive institutions was whether or not they 
settled in large numbers. In colonies where Europeans settled, institutions were developed 
for their own future benefits. In colonies where Europeans did not settle, they often set up 
a highly centralized state apparatus, and other similar institutions, to oppress the native 
population and facilitate the extraction of resources in the short run. Based on this 
idea, Acemoglu, Johnson and Robinson (2001) suggest that in places where the disease 
environments made it easy for Europeans to settle, the path of institutional development 
should have been different from areas where Europeans faced high mortality rates. 

In practice, during the time of colonization, Europeans faced widely different mortality 
rates in colonies because of differences in the prevalence of malaria and yellow fever. 12 
The argument in Acemoglu, Johnson and Robinson (2001) is that differences in mortality 
rates of potential settlers, driven mostly by malaria and yellow fever, provide a possible 
candidate for the variable Z above: these mortality rates should not influence output 
today directly; but by affecting the settlement patterns of Europeans, they may have had 
a first-order effect on institutional development. The idea that these mortality rates should 
not have a direct effect is plausible. Malaria and yellow fever were fatal to Europeans 
who had no immunity, thus having a major effect on settlement patterns, but they had 
much more limited effects on natives who, over centuries, had developed various types of 
immunities. 13 

The data support the notion that there were major differences in the institutional 
development of the high-mortality and low-mortality colonies. Figure 1 shows a measure 
of property rights enforcement — the protection against expropriation risk — against the 
logarithm of potential European settler mortality in 1000 mean strength soldiers (see 
Acemoglu, Johnson and Robinson, 2001 for details). Expropriation risk is much greater 
in places where Europeans faced higher death rates and did not settle. 

12 See Appendix Table A2 in Acemoglu, Johnson, and Robinson (2001) for the variation in the mortality 
rates of European military and clergy in the various colonies. Before 1850, the annual mortality rates for 
a settlement size maintained at 1000 (via replacement) ranged from 8.55 in New Zealand (lower than in 
Europe at that time) to 49 in India, 130 in Jamaica, and around 500 in West Africa. 

13 This "exclusion restriction" is supported by the death rates of natives in these areas. For example, 
Curtin (1964) reports that the annual death rates of native troops serving in Bengal and Madras were 
respectively 11 and 13 in 1000. These numbers are similar to the annual death rates of British troops 
serving in Britain, which were approximately 15 in 1000. In contrast, the death rates of British troops 
serving in these colonies were much higher because of their lack of immunity. For example, death rates 
in Bengal and Madras for British troops were between 70 and 170 in 1000. 



13 



Acemoglu, Johnson and Robinson (2001) also show that these institutional differences 
induced by mortality rates and European settlement patterns have a major effect on in- 
come per capita. 14 For example, the estimates imply that improving Nigeria's institutions 
to the level of those in Chile could, in the long run, lead to as much as a 7-fold increase 
in Nigeria's income. These results have a clear interpretation: in practice, societies do 
choose very different institutions, and not because of differences in output-relevant vari- 
ables, the X's, but because of other political/historical circumstances, the Z's, in this 
case the mortality rates faced by early European settlers. Moreover, these institutional 
choices have a major effect on economic performance. These results suggest that the 
PCT, which emphasizes the forces that push societies towards the correct institutions 
and policies, does not provide a useful framework for analyzing the major institutional 
and policy differences across countries. 

3.3 Another Experiment: North Versus South Korea 

Another example that illustrates how societies with very similar conditions, but different 
histories or political equilibria, may end up with very different economic and political insti- 
tutions, and consequently with divergent economic performances, is the contrast between 
North and South Korea. 

Until the end of World War II, Korea was under Japanese occupation. Korean in- 
dependence came shortly after the Japanese Emperor Hirohito announced the Japanese 
surrender on August 15, 1945. After this date, Soviet forces entered Manchuria and North 
Korea and took over the control of these provinces from the Japanese. The United States 
did not want to leave the control of the Korean peninsula to the Soviet Union, and with 
"General Order No. 1," President Truman proposed a joint occupation of Korea, with 
the division between the north and south at the 38th parallel. The major fear of the 
United States during this time period was the takeover of the entire Korea either by the 
Soviet Union or by communist forces under the control of the former guerrilla fighter, Kim 
II Sung. U.S. authorities therefore supported the influential nationalist leader Syngman 
Rhee, who was in favor of separation rather than a united communist Korea. Elections 

14 That paper also documents that this effect of institutions on economic performance is robust to 
excluding Australia, New Zealand, Canada, and the United States, or Africa, to controlling for various 
geography variables such as distance from the equator (latitude), continent dummies, temperature, hu- 
midity, whether countries are land-locked, soil quality, natural resource abundance etc. They also obtain 
similar results using only yellow fever prevalence, which is an attractive source of variation, since yellow 
fever is now mostly eradicated. 



14 



in the South were held in May 1948, amidst widespread boycott by Koreans opposed to 
separation. The newly elected representatives proceeded to draft a new constitution and 
established the Republic of Korea to the south of the 38th parallel. The North became 
the Democratic People's Republic of Korea, under to control of Kim II Sung. 15 

A distinguishing feature of Korea before separation was its ethnic, linguistic and eco- 
nomic homogeneity. The north and south are inhabited by essentially the same people, 
with the same culture, and there were only minor economic differences between the two 
areas. If anything, at the time of separation, the North was more industrialized; for ex- 
ample, production levels of heavy industrial output were almost four times as high in the 
North as in the South, despite the larger size and population of the South (production of 
light industry in turn was greater in the South, see Ha-cheong, 1988). 

After separation, policies and institutions diverged substantially in the two countries. 
The North, under the leadership of Kim II Sung, adopted a centrahzed command economy 
with httle role for individual enterprise. Kim II Simg advocated and largely imposed a 
philosophy he named "Juche", which played an important role both in the pohtical and 
economic life in North Korea. This philosophy emphasized self-reliance and the control 
of resources by the communist party and the state which, in turn, were supposed to 
represent the people. All non-labor factors of production were under the control of the 
state, which directly made the majority of the key economic decisions. Before the separa- 
tion, industries in North Korea were concentrated in mining, electricity, steel, chemicals, 
transportation, communication and cement. Most of these were quickly nationalized. 
There were also many small household industries and producers, and these were forced 
to join the cooperatives of the Consumer Union, where they were closely supervised and 
instructed by the state. For aU practical purposes, there were no private property rights 
for individuals (see, for example, Koo, 1992, or Eberstadt, 1999). 

In contrast, South Korea, though far from a free-market economy, relied on a capitalist 
organization of the economy, with private ownership of the means of production, and legal 
protection for a range of producers, especially those under the umbrella of the chaebols, 
the large family conglomerates that dominated the South Korean economy. Although 
Syngman Rhee, and subsequently General Park, were highly dictatorial, for a variety of 
reasons beyond the scope of this paper, they refrained from the most predatory policies. 
In fact, General Park was generally supportive of economic development, and his regime is 
15 See Cumings (1997) and Buzo (2002) for recent histories of Korea. 



15 



often credited with facilitating, or even encouraging, investment and rapid growth in Korea 
(e.g., Evans, 1998, Wade, 1990). Even though many South Korean economic policies, such 
as protected domestic markets, entry barriers and subsidized loans, directly favored the 
chaebol, there were no major violations of property rights for the rest of the society, and 
the state actively subsidized and encouraged education. Overall, South Korean economic 
institutions were highly capitalistic, even though the government intervened more than 
the simplest textbook model of free-market capitalism would suggest. 

Under these two highly contrasting regimes, the economies of North and South Korea 
diverged. In 1950, according to Maddison (2001), both North and South Korea had 
approximately the same income level, $770 (in 1990 International Geary-Khamis dollars). 
In the 1990s, before the collapse of the Soviet system and the cessation of Soviet aid, 
Maddison (2001) estimates per capita income in North Korea to be around $2,841, less 
than one-third of the income per capita in South Korea, which stood at $8,704. The South 
Korean government estimates the North Korean GDP per capita to be less than l/6th 
of the South Korean per capita in 1990 (see www.bok.or.kr). In any case, there is no 
doubt that income in North Korea in 1990 was inflated by Soviet aid, and since then, the 
North Korean economy has been shrinking, while South Korea continued to grow rapidly. 
According to Maddison (2001), there is now an over 10-fold difference: income per capita 
is $12,152 in South Korea vs. $1,183 in North Korea. Figure 2 shows the divergence in 
income per capita between South and North Korea using Maddison's (2001) numbers. 

Overall, since 1950, South Korea grew rapidly under capitalist institutions and poli- 
cies, while North Korea experienced minimal growth, under communist institutions and 
policies. This "experiment" of dividing this homogeneous country into two parts with 
very different policies and institutions gives another clear example of how, despite the 
very similar economic conditions, political leaders often choose very different policies, 
with very different outcomes. 

3.4 Conscious Choices or Belief Differences? 

Can we interpret the differences in institutional development across the European colonies 
or the divergence in the institutions and policies between the North and South of Korea 
as resulting from differences in beliefs? For example, could it be the case that while Rhee, 
Park, and other South Korean leaders, believed in the superiority of capitalist institutions 
and private property rights enforcement, Kim II Sung and Communist Party members in 



16 



the North believed that communist policies would be better for the country? 

In the case of South versus North Korea, this is certainly a possibility. However, 
even if differences in beliefs could explain the divergence in institutions in the immediate 
aftermath of separation, by the 1990s or even by the 1980s, it was clear that the communist 
economic policies in the North were not working. The continued efforts of the leadership 
to cling to these policies and to power can only be explained as those leaders wishing to 
look after their own interests at the expense of the interests of the population at large. 
Currently, North Korean leaders, the Communist Party and the bureaucratic elites are 
prolonging the current system, which gives them greater economic and political returns 
than the alternative, while fully realizing the costs that the system imposes on the North 
Korean people, including the famine that much of the population has been suffering over 
the past several years. 

Differences in colonial policies make an even clearer case for the importance of social 
and distributional conflict leading to inefficient policies and persistently inefficient insti- 
tutions. It was the same British colonists who established different institutions in very 
different parts of the world: in the Caribbean and Southern United States, they set up 
plantation societies based on slavery, supported by highly oppressive institutions. In con- 
trast, the institutions they developed in areas where they settled, and where there was no 
large population of Indians or slaves to be oppressed, such as Northeastern United States, 
Canada, Australia and New Zealand, were democratic, encouraged participation, imposed 
checks and balances on politicians and political elites, and generally enforced the property 
rights of a broad cross-section of society. 16 Moreover, differences in the incentives of the 
colonists in various colonies are easy to understand: when they did not settle, they were 
choosing the institutions simply to extract resources from the native population. When 
they settled in large numbers, institutions and policies were set in place in order to protect 
them in the future and encourage investment and prosperity. 

16 An interesting example of how the same groups adopted very different colonization strategies and 
organizations of society in response to different incentives is the experience of the Puritans in the New 
World. While the colony of Massachusetts Bay, formed in 1630 by the English Puritans is hailed as 
an example of good institutions introduced in the colonies by a group seeking economic and religious 
freedom, at the same time a group of Puritans sponsored by the powerful Puritan interests in England 
formed a colony in Providence Island in the Western Caribbean. Slavery was immediately adopted in 
this colony, and the most profitable endeavor for those settling on this island was attacking and pirating 
Spanish ships in the area (see Kupperman, 1988). 



17 



4 Commitment Problems and The Political Coase 
Theorem 

In the previous section, I developed the argument that the PCT, in its simple or modified 
form, does not provide a good framework for thinking about the major cross-country 
differences in institutions and policies. The argument was essentially empirical. In this 
section, I turn to a discussion of why we should expect socially and politically powerful 
groups to often extract resources from the rest of the population in inefficient ways and 
set up bad institutions. In other words, why should we expect the PCT not to hold? 

The basic idea is that the reasoning of the Coase Theorem requires transfers from one 
party to another, and all such transfers cannot be made at the same time. Therefore, 
some type of "enforceable contracts" specifying future transfers are necessary. And yet, 
when such transfers are between the citizens and the state (or groups controlling the 
state), there is a major commitment problem: there is no outside party to enforce such 
contracts, and those controlling the state can renege on their promises. Only "incentive 
compatible" or "self-enforcing" arrangements can be made. These inherent commitment 
problems often rule out the reasoning of the PCT. 

To simplify the discussion, I focus on the case with no belief differences, i.e., the PCT, 
though obviously the discussion here applies to the Modified PCT as well. 

4.1 Description 

Consider the following infinite horizon economy. Time is discrete and indexed by t. There 
are two groups of agents, a ruler, and a mass 1 of identical citizens. All agents discount 
the future with the discount factor /?, so have a utility function 



«t = Yl P t Ct +J _ ( 2 _ a ) e *+j] i 



i=t 



where c t +j is consumption and e t + } is effort (investment), and the term (1 — a) is intro- 
duced to simplify some of the algebra below. 



yl = f(e\) = {e\) i - a + R 

There is also another inferior production technology, which has the advantage of being 
non-taxable. In particular, this alternative produces nonmarket income 



18 



where b < 1. I denote the decision to produce for the market by m t G {0, 1}, and when 
m t = and the individual uses the non-market technology, his market income is y\ = R, 
so income from natural resources, R, remains taxable. 17 

In the first-best — efficient — allocation, only the superior market technology would be 
used, and the level of investment would satisfy: 

de t 

There is a ruler with the power to tax the citizens. The ruler does not contribute to 
the production process, but because he has all the means of coercion, he can take as much 
of the output in the market sector as he wishes. This clearly ignores useful roles of rulers 
and of the state, such as law enforcement, public good provision, regulation and defense. 
This is only for simplicity, since these roles are not essential for the theory here. 18 

The feasibility constraint that determines the maximum tax per person that the ruler 
can impose is 

T t (Y)<Y= [yldi, (1) 



where Y denotes aggregate output. In addition, the taxation technology needs to be 
specified. The most general case would be when the ruler specifies the person-specific: 
taxes, T? , for each individual j. I return to a discussion of this issue below. 

Rulers have the same discount factor, /?, as the citizens, but because of internal power 
struggle, they can also lose power to another ruler with exogenous probability q (see 
Acemoglu, 2002, for a similar model where this probability is endogenized). The current 
ruler can also decide to relinquish his power, in which case there are no more rulers in the 
future, and nobody can tax the citizens. I refer to this case loosely as "democratization," 
though it lacks many of the interesting features of real-life experiences of transitions to 
democracy (see Acemoglu and Robinson, 2000a, for a model of democratization). I will 
make two alternative assumptions regarding what type of contracts can be enforced: (i) 
contracts that citizens write with a current or previous ruler can be, at least partly, 

17 The presence of market income, even when individuals withdraw from market production will ensure 
that rulers continue to get positive return. 

18 See Acemoglu and Verdier (1998) for a model where government plays a useful role, but also govern- 
ment officials are corrupt and their actions distort private incentives. 



I't 



"enforced"; (ii) no enforcement of such contracts. The plausible case is clearly (ii), but 
(i) is useful to analyze as a benchmark. 

The timing of events within each period is as follows: 

1 . If contracts are available, parties sign contracts. If there has been no democratiza- 
tion in the past, i.e., r = in all past periods, then the ruler decides whether to 
relinquish his power, r = or r = 1. 

2. Individuals choose how much to invest, e, and whether or not to produce in the 
market sector, m = or 1 . 

3. If r = in all previous periods, the ruler decides how much aggregate tax T (Y)' 
to impose on the citizens, as a function of aggregate income Y. If r = 1 in some 
previous period so that there is "democracy" , then a citizen decides how much tax 
S (Y) to impose on each individual to be given to a previous ruler. 

4. Consumption takes place. 

5. If there has been no democratization, it is revealed whether or not the ruler will be 
in power in the next period (he is replaced with probability q). 

This timing of events introduces the assumption that not all transactions can be 

made at the same time; citizens invest first, and rulers set taxes after. So some type 

of "contracts", implicit or explicit, are necessary. 19 The history of play in this repeated 

game, h*, includes all the actions up to that point. The strategy of a ruler consists of a 

mapping a (■ | h f ), which determines (r, T (Y)) in every period for a given history h l . The 

level of taxes T is in turn conditioned on the level of output, since, according to the timing 

of events, taxes are determined after citizens make their investment and sector choices, 

and also because taxes can never exceed the level of output. The strategy of citizens 

consists of a mapping p (■ | h 1 ), which determines (m, e, S (Y)) for a given history of the 

game h l (I am focusing on symmetric equilibria where all citizens use the same strategy, 

hence I specify only one strategy mapping for the citizens, p (■ | h*)). The investment and 

19 This game also introduces a possible distinction between institutions and policies: institutions may 
correspond to whether the society is democratic, i.e., who has the right set taxes, while policies corre- 
spond to the choice of actual taxes. Nevertheless, my focus here is not to clarify the distinction between 
institutions and policies, but highlight the forces that prevent the efficient choice of policies (and institu- 
tions). 



20 



sector choices of citizens are conditioned on the actions of the ruler in the same period 
that are observed before the citizens' actions. 

A subgame perfect equilibrium is defined as a strategy a (■ | h l ) for the ruler and a 
strategy p (■ | h l ) for all citizens that are best responses to each other in all subgames, i.e., 
for all h*. To simplify the discussion throughout, I focus on stationary equilibria, where 
the same strategies are played at all dates. 

4.2 The No-Cooperation Benchmark 

Let us start with "no-cooperation benchmark," which features r = 0, i.e., no democrati- 
zation, and no contracts between rulers and citizens. 

Proposition 1 Suppose 7- = 0. Then there exists an equilibrium in which all citizens 
expect the ruler to grab everything, so they use the informal sector technology, m = 
and e = b. The ruler sets T (Y) = Y. 

It is straightforward to see that this allocation is an equilibrium. It is a weakly dom- 
inant strategy for the ruler to grab everything, which along the equilibrium path will 
simply be the income from natural resources, R. And if an individual deviates and pro- 
duces more, this will not increase his consumption, since the ruler is grabbing everything. 
So the citizens choose m = 0, i.e., production with the nomnarket technology, and they 
invest the optimal amount for this technology, e = b. 

For future reference, denote the values of the citizens and the ruler in this equilibrium 
as W and V: 

*-!#*• < 2 » 

and 

9 -rrWTY (3) 

This equilibrium is highly inefficient. For example, a contract along the following lines 
would constitute a Pareto improvement: the ruler relinquishes power, and the citizens 
promise him a side payment of R + e every period thereafter. Then, they would all choose 
market production, e = 1, achieving the first-best equilibrium. The focus below will be 
on whether this type of arrangement can be made to improve the allocation of resources 
away from that specified in Proposition 1. 



21 



4.3 The Political Coase Theorem With Commitment 

Next I discuss the equilibrium allocation when enforceable contracts between rulers and 
citizens are possible. In this discussion, I will say that a Political Coase Theorem (PCT) 
applies when, even in the absence of full property rights for citizens, the economy generates 
the same allocation (and when the distribution of political power between the citizens and 
the ruler is irrelevant for the allocation). 

There are three different ways in which this result may emerge: with full commitment 
on the side of the ruler, with full commitment on the side of citizens, and with limited 
commitment. In this subsection, I discuss the first two cases, leaving the third, which is 
the central focus of this analysis, to the next section. 

First suppose that the ruler can commit to impose a tax level T during this period 
(more specifically, T (Y) = min (T; Y)), and assume that q = 0, so that there is no ruler 
replacement. This means that each citizen will pay a tax level T. Then, each citizen knows 
that he will be able to keep any level of production above T. The following program gives 
the equilibrium allocation that satisfies the PCT, yielding the largest surplus to the ruler, 
with r = (i.e., no "democratization"): 

T 

max -, 

T,e 1 -/?' 

subject to the feasibility constraint (1), and an incentive compatibility constraint for 
citizens: 

W{e) = j^— j [e 1 - Q -{l-a)e + R-T)>W, (4) 

where the left-hand side of (4) is the return to citizens when they invest e, and are 
taxed T, and the right-hand side, W, is the value that citizens can obtain opting out of 
the formal sector, given by (2). The solution to this problem is straightforward: T = 
a (1 - b) + R (or more formally, T (Y) = min (a (1 - b) + R; Y)) and r = for the 
ruler, and e = 1 and m = 1 for all the citizens. The important point is that the efficient 
allocation is achieved despite the fact that political power is in the hands of the ruler. The 
reason for this is the commitment power of the ruler: by committing to the tax schedule 
T (Y) = min (a (1 — b) ; Y), the ruler is making the citizens the residual claimant after a 
certain level of investment, and this encourages them to undertake the first-best level of 
investment. 

The above program was special in that it gave all the "bargaining power" to the ruler. 
Alternatively, we can have some of the rents from achieving the PCT shared between the 

22 



ruler and the citizens. The simplest way to illustrate this is to assume that rents between 
citizens and rulers are shared by Generalized Nash Bargaining. By the same reasoning as 
above, citizens will choose the efficient level of investment, e = 1, so imposing this level 
of investment, the Generalized Nash Bargaining solution can be found from the following 
program: 

1-0 

(5) 



max 
T 



a R T ab 

l-/3 + l-/? _ l-/?~l-/3 



T R 



1-/3 1-/3 



subject to (4), where 9 is the bargaining power of the citizens. The first bracket is 
the return to citizens net of their outside option, which is production for the non- 
market sector with net present value, ab/ (1 — /3). The second bracket is the net re- 
turn to the ruler above his outside option of taxing, only the income from natural re- 
sources. The solution to this problem has T = (1 — 8) a ■ (1 — b) + R (or more formally, 
T (Y) = min ((1 - 8) a : (1 - b) 4- R; Y)) and r = for the ruler, and e = 1 and m = 1 
(production in the market sector) for all citizens. Note that the surplus from citizens pro- 
ducing in the market sector and undertaking the first-best level of investment, (over the 
alternative of non-market production) is a (1 — b). This surplus is being shared between 
the citizens and the ruler. Income from "natural resources," R, goes entirely to the ruler, 
since the ruler can obtain this even when citizens do not "cooperate" . As the bargaining 
power of citizens, 9, goes to zero, we obtain the tax level above, T = a (1 — b) + R . 
Once again, the important point is that the efficient allocation is achieved, thanks to the 
commitment power of the ruler. Moreover, the model also illustrates that the distribution 
of political power between the ruler and the citizens, 9, does not affect the efficiency of 
the allocation; m = 1 and e = 1 irrespective of 9. 

This solution is slightly more involved when the ruler can be replaced by a new ruler, 
i.e., q > 0. In this case, rulers have a preference for front-loaded payments, since they may 
not be be around in the future, i.e., they discount the future at the rate (3 (1 — q), which is 
less than the discount factor of citizens, (3. However, citizens dislike making front-loaded 
payments to current rulers, since in case these rulers are replaced, these payments are 
lost, and additional payments have to be made to new rulers. These two effects cancel 
each other, and the problem is still stationary. In particular, now the allocation will be a 
solution to maximization problem: 

9 



max 

T 



a R T ab 

+ 



1-/3 1-/3 1-/3 1-0 

23 



T R 



[l-/3(l- g ) l-/3(l-g)J 



subject to (4). This only differs from (5) because the discount factor of the ruler is different 
due to potential replacement at the end of the period. The solution is straightforward to 
characterize, and is identical to above. With complete contracts, the discoimt factor of 
the ruler does not matter for the equilibrium allocation. 

Next suppose that the ruler cannot commit to a tax level T, but citizens can commit 
to a future path of transfers, {S t } if the ruler relinquishes his power. Now the PCT can 
be achieved via democratization, i.e., r = 1; the ruler transfers power to the citizens in 
return for their commitment to a future path of transfers. The equilibrium allocation with 
commitment on the side of citizens is therefore similarly a solution to the maximization 
problem: 

1-0 



max 

s 



a R S ab 

+ 



1 -0 1-0 1-0 1-0 



S R 



[l-0(l-q) l-0(l-q)\ 



subject to (4). The solution is: S (Y) = min ((1 - 9) a (1 - b) + R; Y)) and r = 1 for 
the ruler, and e = 1 and m = 1 for all citizens. Therefore, with commitment to future 
taxes and transfers on the side either the ruler or the citizens, the basic logic of the 
PCT is powerful, and the distribution of rents between various parties (here the ruler and 
the citizens) can be separated from efficiency considerations. The first-best investment 
level is achieved, and the distribution of power, here captured by 9, has no effect on the 
allocation. 

Proposition 2 When either rulers or citizens can commit to future transfers, we always 
have m = 1 and e = 1, i.e., the PCT applies. 

4.4 Limited Commitment 

The above analysis with "contracts" between rulers and citizens is useful as a benchmark, 
but has little practical relevance. These types of contracts are not enforceable in the real 
world. Contract enforcement requires a third party, typically the state, which possesses 
the monopoly of legitimate coercion in the society. Its monopoly of coercion gives the 
state the power to force contractors to abide by the terms of the contract, even if ex post 
making the specified payments or the necessary delivery of goods is not in their interests. 
When the state is one of the "contractors", this type of outside enforcement is not 
possible. For this reason, it is very difficult for any party with real power to commit to a 



24 



path of future transfers, taxes or actions. Therefore, we cannot hope for the outside en- 
forcement when we are dealing with interactions with the state; abiding by the conditions 
of the "contract" has to be incentive compatible for the state (and for the citizens). 

To highlight these issues, I now analyze the above game without such contracts. I start 
with the Markov Perfect Equilibria (MPE) of this game, which does not allow repeated- 
game punishment strategies. In the next section, I will analyze non-Markovian equilibria. 

Using the above notation, in this simple game, an MPE is defined as a strategy combi- 
nation a (■ | /?') for the ruler and p (• | h l ) for the citizens that are best responses to each 
other, and history-independent, i.e., a (• | h l ) = a (■ | h n ) and p(- \ h') = p(- | h n ) for any 
h l and h n . Thus strategies in a MPE do not depend on the history of the game (more 
generally, they depend only on the payoff-relevant state variables, and here there are no 
such state variables). 

This implies that within each period, we can solve the game by backward induction. 
In the last stage, the ruler in power sets the tax. The best action for the ruler is to grab 
everything, since the future play of the game, and therefore the continuation payoffs, does 
not depend on history, so whether or not he has grabbed everything will not have future 
repercussions. Hence, T (Y) = Y. Given this, citizens prefer m = 0, and there is no 
market production. We are back to Proposition 1. 

This is a highly inefficient outcome, and one that both citizens and the ruler would 
like to prevent. For example, the ruler would like to promise to set a lower tax, e.g., 
T (Y) = min (T; Y) for some T < a (1 — b) + R, which would encourage citizens to stay in 
the market and invest up to the first-best level of investment. However, no such promises 
can be credible (without some type of trigger punishment strategies). The PCT, therefore, 
does not apply because of lack of commitment. 

Proposition 3 Without the possibility of commitment by the ruler or the citizens to 
future actions, the unique Markov Perfect Equilibrium has m = 0, and T (Y) = Y. 

5 Incentive- Compatible Promises 

5.1 Incentive-Compatible Commitments by the Ruler 

The discussion with MPEs ignored possible commitment that can be supported thanks 
to the repeated nature of the game. For example, if we allow strategies to depend on 
the history of the game, the citizens and the ruler may enter into an implicit agreement 

25 



where the ruler promises not to grab everything because of future rents from continued 
market production by the citizens. The important point is that such promises have to 
be "self-enforcing" or incentive compatible for the ruler. We can capture these issues by 
analyzing the non-Markovian equilibria of tins game, where citizens play trigger strategies 
to induce the ruler not to grab all the output. 

I will undertake this analysis in stages, starting with the case where there is no re- 
placement of rulers, i.e., q = 0. Moreover, I first assume that the citizens can coordinate 
their actions and all choose the level of investment e, maximizing their utility as a group. 
Thus we can think of the game as one between two players. Later in subsection 5.3, I 
will come back to the issues of "free-riding," where each individual may prefer to choose 
a different level of investment than the one maximizing the utility of the citizens as a 
group. These considerations will have interesting implications for the equilibrium form of 
taxation. 

Consider the following strategy combination for the ruler and the citizens: the ruler 

sets the tax T (Y) = min (T; Y) as long as citizens have played e' = e in all past periods, 

and T (Y) = Y otherwise; citizens play m = 1 and e' = e as long as the ruler has set the 

tax T (Y) = min (T; Y) in all past periods, and m = otherwise, and we have T < e a + R. 

The resulting allocation will yield a tax revenue of T in each period, and give the ruler 

utility: 

T 

v =—0- (6 > 

Since the ruler cannot commit to future taxes of the form T (Y) = min (T; Y) , we 
have to ensure that playing the above specified strategy is optimal for the ruler. The 
obvious, and the best, deviation for the ruler from this strategy profile is clearly to grab 
everything in the current period. So we have to check that not deviating from this 
strategy profile (i.e., not to grab everything today) is incentive compatible. If the ruler 
follows the repeated game equilibrium, he obtains V as given by (6). Alternatively, if he 
deviates to grab everything today and then switches to the non-cooperative equilibrium 
of Proposition 1, he obtains all the output today, e 1_Q + R, and from today onwards, 
he obtains his payoff in the non-cooperative equilibrium of Proposition 1, V. Thus the 
return to deviating is e 1_Q + R + (3V . 

Incentive compatibility for the ruler therefore requires: 

e l ~ a + R + (3V <V, 



26 



or written more compactly, the incentive compatibility constraint for rulers is 

T>T{e) = {l-f3)e l - a + R, (7) 

where the function T (e) is defined for future reference, and represents the flow value of 
grabbing all current output for the ruler when current investment is e. Condition (7) 
states that the tax that the ruler receives in each period should be large enough so that 
he is not tempted to grab everything. 

We also have to satisfy incentive compatibility for citizens, to convince them to stay 
in the market sector. When they do so, they obtain 

w ( e ) = T^]3 [ el ~ Q ~ (1 ~ a) e + R ~ T l ' (8) 

which needs to be greater than W given by (2) for the "equilibrium" e. In other words: 

T <T m ^{e) = R + e l ~ a -{l-a)e-ab. (9) 

Here T max (e) is the maximum tax that citizens are willing to pay before they switch to 
the non-market sector. 

I now look for an equilibrium which satisfies these two incentive compatibility con- 
straints. First, I will check whether the first best can be supported, i.e., whether the 
allocation with m = 1 and e = 1 can be achieved, so that the basic insights of the PCT 
generalize to this case without commitment. 

To investigate the conditions under which the first-best allocation with e = 1 can be 
supported, observe that the maximum tax rate consistent with citizens incentive compat- 
ibility constraint is 

r max (e = l)-H + a(l-6). 

Then the question of whether we can support the first-best allocation simply boils down 
to whether we can satisfy the ruler's incentive compatibility constraint, (7) with this tax 
level. Combining this with (7), we have: 

T max (e = 1) = R+ a (1 - b) > T (e = 1) = 1 - p + R, 

which is equivalent to the condition: 

1-j9<o(1-6) (10) 



27 



When (10) is satisfied, the basic message of the PCT goes through. Agents can enter into 
implicit agreements because the threat of punishment implied by the trigger strategies is 
sufficient to overcome the inherent commitment problems, and the first-best allocation 
can be achieved despite the fact that all political power is in the hands of the ruler. 
Condition (10) is more likely to be satisfied when agents are patient, and the outside 
options of citizens are not too attractive, so that the ruler can raise enough taxes in every 
period not to be tempted to tax more than the prescribed amount. 

Next consider the case where (10) is not satisfied, so that the first-best investment 
level, e — 1 , cannot be maintained. Can market participation by the citizens, m = 1 and 
some positive investment in the market sector, e > 0, be supported as an equilibrium? The 
answer is yes provided that we can satisfy the inequality that the maximum tax citizens 
are willing to pay is greater than the flow return to the ruler from grabbing everything: 

T max {e) = R + e 1 '" - (1 - a) e - ab > T (e) = (1 - 0) e l ~ a + R, (11) 

The left-hand side, r max (e), represents the incentive compatibility condition of the citi- 
zens, while the right-hand side, T (e), corresponds to the incentive compatibility condition 
of the ruler. 

Figure 3 draws the left- and right-hand sides of this equation in the space of e 1-Q and T. 
For low values of e, T™ 1 ** (e) increases faster than T (e), so greater investment levels make 
it easier to satisfy both incentive compatibility conditions. However, the gap between 
T" 1 ^ (e) and T (e) reaches its maximum at e = (5 l ' a < 1, where the slopes of the two 
curves are equalized. After this point, T (e) grows faster than J 1 ™ 3 * (e). This reflects the 
fact that the incentive compatibility constraint of the ruler depends on output, whereas 
the incentive compatibility of the citizens depends on the difference between output and 
the cost of investment, which grows less than output. 

As a result, we always have that 7™* (e = (3 1/a ) - T (e = P 1/a ) > T max (e = 1) - 
T (e = 1), and it is easier to satisfy both incentive compatibility constraints at e = l ' a 
than at the first-best level of investment. This reflects the fact that at maximum effort, 
there are strong incentives for the ruler to grab everything today. 

Figure 3 shows a case where T raax (e = (3 1/a ) - T (e = (5 l/a \ > > T max (e = 1) - 
T (e = 1), so while the first-best cannot be supported, a range of investment levels, e e 
[e*,e**], can be supported as an equilibrium of the repeated game between the ruler and 
the citizens. As we will see in greater detail below, an equilibrium with e = /3 1 ' is 
preferred to any equilibrium with e € [e*,/? 1/,Q ) by both the citizens and the ruler. So 

28 



we can focus on the set e € 



/3 17 V 



at the set of potential equilibria. It is clear that 
this set changes with the underlying parameters. For example, when (5 increases, the 
set becomes larger, and in particular, the highest investment that can be supported, e**, 
increases (Figure 3, in fact, shows the case where, with the decline in /?, the first-best 
level of investment can be supported ) . 

This analysis also gives us a simple condition to check to see whether market produc- 
tion can be supported. If we cannot satisfy both incentive compatibility constraints when 
e = /? 1/,Q , then the set [e*,e**] will be empty. Therefore, the condition for m = 1 to be 
supported is 

a0 1/a > b (12) 

Now suppose that condition (12) is satisfied, so that the set [e*,e**] is non-empty. 
Which of the many equilibria, or which of the investment levels in the set [e*, e**], will be 
chosen? This paper has little to say on equilibrium selection, so any of these investments, 
as well as many others that are Pareto inferior to those on the frontier, may emerge in 
equilibrium. Nevertheless, it is useful to briefly discuss which of these equilibria will be 
the most preferred by the citizens and by the ruler, and how changes in the distribution 
of "power" between the two groups might affect equilibrium selection. 

First, consider maximizing the ruler's utility, (6) subject to (7) and (9) by choosing 
e and T. The solution to this problem is e = e** and T max (e**): the ruler would like 
to maximize investment and choose the highest possible tax level given that investment. 
Incidentally, this is exactly what a social planner who wishes to maximize output would 
also choose. 

In contrast, citizens would like to maximize (8) again subject to (7) and (9). We 
know that as long as e e [e*,e**], citizens' incentive compatibility constraint, (9), will be 
satisfied, and citizens will never give the ruler more than the minimum amount to satisfy 
his incentive compatibility. Therefore, the ruler's incentive productivity constraint, (7), 
has to hold as an equality. A simple way to characterize the solution is then simply to 
take this equation and substitute it back into (8). This gives the maximization problem: 

maxT max (e) - T (e) = (3e l ~ a - (1 - a) e, 

i.e., citizens would like to maximize the difference between the left-hand and the right- 
hand sides of (11), which has the solution e = (3 l ^ a . Intuitively, increasing effort further is 
costly for the citizens, because they pay the additional investment costs, while the ruler 

29 



gets the benefits. Since they do not internalize the ruler's gains, they prefer e = /3 1 / to 
the maximum investment that can be supported. 

The Generalized Nash Bargaining solution between the ruler and the citizens is, in 
turn, given by: 



max 

e,T 



1 , „ (l-a)e R T ab 



1-/3 1-/3 1-/3 1-/3 1-/3 



T R 



l-P 1-/3 



subject to the two incentive compatibility constraints, (7) and (9), where as before 9 is 

the bargaining power of the citizens. According to the PCT, the allocation of political 

power between the two groups should not matter for the outcome. Here, it will clearly 

matter (as long as (12) is satisfied, and the set [e*, e**} is non-empty). We already saw that 

when = 0, i.e., when the ruler has all the bargaining power, we obtain e = e**, whereas 

when 9 = 1, we obtain the citizens' most preferred solution, e = /3 1 . Analysis of this 

maximization problem establishes that the general solution is e (9), which is decreasing in 

9 with e (9 = 0) = e** and e (9 = 1) = /? 1//q . 20 That greater bargaining power for citizens 

reduces investment and efficiency is a somewhat surprising result, but is intuitive ex post. 

Recall that the problem is the inability of the ruler to commit to not taxing the returns 

from citizens' investments; so a naive intuition may have been that greater bargaining 

power for the citizens would reduce inefficiencies. However, the bargaining power 9 does 

not affect the incentive compatibility constraint of rulers — instead, it determines which 

point we choose from the possible set of allocations. Since citizens bear the cost of 

investment and receive less than the full return, their preferred investment is less than 

that of the ruler. Greater bargaining power for the citizens selects an equilibrium closer 

to their desired point, with lower investment and greater net returns for them. 

Next consider the case with ruler replacement, i.e., q > 0. Since replacement happens 

at the end of the period, the only difference from the above analysis is the value of 

20 To see this, first note that the incentive compatibility constraint of the ruler, (7) has to hold (oth- 
erwise, both parties can be made better off). Using this condition and factoring out constants, the 
maximization problem can be rewritten as 

max \(3e l - a - (1 - a) e - ab] 9 [e 1 " ] 1 "" 
Differentiating and simplifying, we obtain 

(e — b) a 
which gives de/d9 < in the range e 6 [e (0 = 1) = 1/a , e (0 = 0) = e** 

30 



continued cooperation for the ruler. Taking this into account, the relevant comparison for 
the ruler is between grabbing everything, which has payoff e 1_Q + Rj (1 — (1 — g)), and 
taxing at the prescribed rate, which yields Tj (1 — (1 — q)). Both of these are different 
from the above expressions because the value of the future is less for the ruler due to the 
possibility of replacement. The ruler's incentive compatibility constraint then becomes: 

T>T(e) = (l-p(l-q))e 1 - a + R, (13) 

and citizens' incentive compatibility constraint, remains unchanged. This implies that the 
first-best can now be supported when 

l-0(l-q)<a(l-b), (14) 

wliich is more restrictive than (10) for all q > 0. Intuitively, the possibility of replacement 
reduces the value of future cooperation for the ruler, and makes the first-best and the 
PCT more difficult to achieve. 

The general solution also changes in the same direction. In terms of Figure 3, the curve 
for T (e) shifts up, and the range of investment levels that can be supported declines. The 
bargaining solution now corresponds to maximizing: 



max 
e,T 



1 , „ (l-a)e R T ab 



T R 



1-/9(1-9) l-0(l-q)\ 



1 - 1-/3 1-0 1-0 1-0 

subject to (9) and (13). It is straightforward to show that a greater q, i.e., a higher 
probability of replacement, reduces the equilibrium level of investment. 21 This effect of 
the replacement probability, q, contrasts with the case with enforceable contracts, where 
q did not matter. 

Finally, the corresponding condition for an equilibrium with m = 1 to be supported 
changes to 

(0 (1 - q)) 1/a > b. (15) 

Summarizing: 

21 The mathematical argument mirrors that of footnote 20, and now the relevant expression is 

(e — b) a 
which gives de/dq < in the range e e [e (6 = 1) = /3 1/a , e (9 = 0) = e"| . 



:w 



Proposition 4 When rulers and citizens cannot commit to future transfers, the PCT 
and the efficient allocation can be supported by trigger punishment strategies provided 
that (14) is satisfied. Otherwise, the level of investment is less than the first best, e = 1. 
As long as condition (15) is satisfied, an equilibrium with market production, i.e., m = 1, 
but e < 1, can be supported. In this equilibrium, the level of investment is a decreasing 
function of the bargaining power of the citizens, 9, and of the replacement probability of 
the rulers, q. 

This analysis therefore establishes limits on the reasoning of the PCT because of 
the inherent commitment problems in politics. Since there is no outside party with the 
coercion capability to enforce contracts between rulers and citizens, promises of rulers have 
to be self-enforcing or incentive-compatible. This puts limits on the society's capacity 
to achieve efficient allocations, and on the applicability of the PCT. This is so, even 
though the model does not rule out any types of transfers between citizens and rulers on 
technological grounds. 

5.2 Determinants of Policies and Institutions 

The analysis and the comparative statics above provide us with a simple interpretation of 
the potential determinants of equilibrium institutions and policies. First, the distribution 
of political power between rulers and citizens, 9, matters for the equilibrium outcome when 
the PCT does not apply. More interesting, the horizon of the ruler matters. When rulers 
are impatient because they fear replacement by other competing rulers, self-enforcing 
agreements are harder to maintain because of standard repeated game reasoning. There- 
fore, better equilibrium policies will arise when rulers have longer horizons. 22 Finally, 
better outside options for citizens (leaving only a small surplus to be shared between 
rulers and citizens in the market) make cooperation between citizens and rulers more 
difficult. 

Who designs the game, or the "constitution", may also be important, especially in 
thinking about institutions imposed on a society by external groups, such as colonial 
powers. If the political system is set up by the citizens, they will immediately choose 
r = 1, i.e., democratic institutions, where rulers do not have the power to tax them. 

22 See Acemoglu and Robinson (2000b, 2002) for a different argument for why rulers who fear replace- 
ment may pursue the wrong policies for the society. There, rulers who fear replacement are more likely 
to resist the introduction of superior technologies or institutions when these changes can erode part of 
their incumbency advantage and potential future political power. 

32 



In contrast, if the original design is by the ruler, or by some political elite, who does 
not internalize the interests of the citizens, they will opt for 7* = 0. Though trivial, this 
observation may be important in thinking about why the European colonists introduced 
relatively democratic institutions with checks and balances on state and politicians' power 
in colonies where they settled in large numbers (i.e., where they became the "citizens" in 
terms of the model above), while establishing or maintaining oppressive and extractive 
institutions in colonies where they did not settle and wished to transfer resources from 
the native population to themselves. 

Although the focus of this paper is not to construct a model that can be used to 
interpret a wide range of social situations, it is also instructive to attempt to incorporate 
"checks and balances" in this simple framework. To do this, I now extend the model 
in one dimension to introduce a measure of institutional controls on politicians: costly 
replacement of rulers. This analysis is useful both for providing comparative statics with 
respect to the extent of checks and balances, and to show the interaction between these 
types of institutional constraints on rulers and the implicit constraints that the rulers 
place on themselves via self-enforcing agreements. 

The only difference from the baseline model is that I now assume that after the taxes 
are set, citizens can attempt to replace the politician, but such replacement costs c (in 
terms of the timing of events in subsection 4.1, the replacement decision is after step 3). I 
assume that this cost is incurred by all the citizens irrespective of whether they "support" 
the replacement of the ruler, so there is no "free-rider" problem. 23 If citizens attempt to 
replace the ruler, this is successful with probability p. The parameter p can be interpreted 
as a measure of the quality of checks and balances on politicians: when p is high, citizens 
can control the ruler better. 

If the current ruler is successfully ousted from power, a new ruler is put in place 
the following period. I assume that if the ruler is ousted, he does not receive the tax 
revenue from the current period, and to simplify the analysis I also assume that this tax 
revenue is not received by the citizens either. Similar results are obtained with alternative 
assumptions, but the current set of assumptions simplifies the analysis. 

Now suppose the ruler has set the tax T, and is expected to set the same tax in the 

future. Let us first ignore the incentive compatibility constraint of the ruler, and suppose 

that the same equilibrium will be played over time irrespective of whether citizens have 

23 This refers to free-riding in the decision whether to oust the current ruler, different from "free-riding" 
in the investment decisions discussed in the next subsection. 



33 



attempted to replace, or have replaced, the ruler or not. Also to further simplify the 
discussion, set q = 0. 

The citizens have a choice of whether to replace the ruler, at cost c, or go along with 
the implicit agreement. The value function of citizens at this point is: 

u/( ^ j e l - a -(l-a)e + R-c-T + (3W(e,T)- \ , . 

W(e,T) = maxj e iV_ (1 J a)e + # _ r + £ W (e >) }■ (16) 

The upper branch corresponds to the choice to replace, and the lower branch applies when 
citizens do not attempt to replace. Notice that the continuation value with or without 
replacement is the same, (5W (e, T), since there will be a ruler following the optimal policy 
after this point, and we are assuming that this new ruler will play the same strategy. The 
only difference between the two branches is the cost of replacement, — c, immediately 
implying that citizens will never exercise their option to replace the ruler. It is costly, 
and along the equilibrium path, it creates no benefits. 24 

Nevertheless, the option to replace may have an effect on the equilibrium because 
it may be beneficial for the citizens to replace a ruler who has deviated. In particular, 
consider a ruler who deviates and grabs all the output. Following this, the citizens and 
the ruler will play the no-cooperation game. So the citizens' continuation value, if they 
do not attempt to replace the ruler, is 

W(e) = -(l-a)e + 0W, (17) 

where W is the value of the citizens and the no-cooperation continuation game given by 
(2), and — (1 — a) e is the flow return this period, since they have invested e and all the 
output is grabbed by the ruler. This expression also incorporates the fact that if citizens 
do not replace the ruler now, they will not replace him in any of the following dates. 

Next consider the value to the citizens after they attempt to replace the ruler (and 
reverting back to never replacing Mm after this period, if this attempt does not work), 
presuming that in the continuation game, citizens will cooperate with a new ruler: 

W (e) = - (1 - a) e - c + /3 (1 -p) W + f3pW (e, T) , (18) 

where W (e,T) is the equilibrium value. Comparing (17) and (18), we see that as long as 

c<p(3(w-W(e,TJ) , (19) 

24 This conclusion holds a fortiori if, following an unsuccessful replacement attempt, the ruler and the 
citizens revert to no-cooperation. 

34 



citizens will attempt to replace the ruler. Since, by construction, W — W (e, T) > 0, 
condition, (19), implies that for sufficiently low costs of replacement, i.e., for c — > 0, 
citizens will attempt to replace rulers who deviate and grab all the output. Citizens' 
replacement option will then affect the incentive compatibility constraint of the ruler. In 
particular, when (19) holds, a ruler who deviates and grabs everything knows that he 
will be replaced with probability p. Since W — W {e,T) > also in all future dates, the 
citizens will attempt to replace the ruler in every future period as well. Taking this into 
account, the ruler's incentive compatibility constraint changes from (7) to 



e'- a + 



T>(l-0)(l-p) 
Thus, the condition for the PCT to hold becomes 

(l-p)(l-/?)-a(l-6)< 



R 



l-P(l-p) 



pR 



(20) 



i-(i- p )/r 

which boils down to (10) when p = 0. This condition is more likely to hold when p is 
high. Therefore, better checks and balances on rulers, as captured by a higher value of p, 
make PCT more likely to hold. 

When (20) does not hold, the allocation most preferred by the ruler will be given 
by the maximum investment that satisfies both the citizens's and the ruler's incentive 
compatibility constraints, thus by the maximum e that satisfies: 

< T max (e = 1) = e l ~ Q + R-{l-a)e-ab, 



(l-p)(l-/3) 


«- +1 . 


R 

-d-p)p\ 


or by e such that: 




,1 P R -Ml 


1 n\ Cl 



(1 - p) (1 - /?)] e l ~ a -{l-a)e = ab- 



pR 



(21) 



l-(l-p)/3 L " y l_(l_ p)/? - 

It is straightforward to check that the solution to this equation, e, is increasing in p. 
Therefore, a better "technology" for citizens to replace the ruler will lead to greater 
equilibrium investments. A similar argument to the one before establishes that market 
production, m = 1, can be supported in this case as long as: 



[1-(1-P)(1 

Summarizing this discussion: 



m 



>b- 



a[l-{l-p)ft 



(22) 



35 



Proposition 5 Consider the game with replacement in this subsection, and suppose that 
the cost of replacement, c, small, i.e., c — » 0. Then, if (20) holds, the PCT applies and the 
efficient level of investment can be supported. Better checks and balances, as captured by 
greater p, make (20) more likely to hold. When (20) does not hold, the efficient level of 
investment cannot be achieved. But as long as (22) holds, market production, m = 1, can 
be supported, and in this case, equilibrium investment is given by e that satisfies (21). 
Better checks and balances, i.e., greater p, increase equilibrium investment. 

5.3 Free-Riding, Overinvestment and the Form of Taxation 

The simple model discussed above also raises another set of interesting issues related to 
free-riding among the citizens, and the form of taxation. Once we relax this assumption, 
the model delivers a motivation for distortionary taxation. 

Suppose that citizens do not coordinate their actions, and the ruler, like before, ob- 
serves the aggregate income level, Y, and sets a lump-sum tax T that applies to each 
individual. Let us focus on the case where (14) does not hold but (12) holds. This implies 
that the first-best is not possible, but there exist equilibria with ra = 1, market produc- 
tion, which feature e < 1. Now each individual is facing a lump-sum tax T, and since 
individuals are atomistic, they do not take their effect on Y into account. This immedi- 
ately implies that the equilibrium with e < 1 is no longer possible. Each individual would 
like to invest up to e = 1, since each is infinitesimal and is, at the margin, the residual 
claimant of the returns from the additional investment. This behavior by all individuals 
will take aggregate output to Y = 1 + R, violating the incentive compatibility constraint 
of the ruler, and destroying the equilibrium with market production. 

Is there a way to prevent this type of unraveling of the self-enforcing equilibrium? The 
answer is yes: move away from lump-sum taxation. As long as simple tax schedules con- 
ditional on individual income, y jy are possible, the equilibrium tax schedule can be made 
sufficiently distortionary to induce exactly the right level of investment. For example, 
imagine now that the ruler sets the following linear tax schedule Tj (yj) = tq + T\yj. In 
response, investment in the market sector would be e = (1 — Tj.) 1 . Suppose that the 
equilibrium that we would like to support has e < 1 and tax level T. Then in order to 
be able to support this equilibrium, and the ruler has to set the following tax schedule 
e = (1 — 7~i) and tq = T — T\ (1 — T\y ~ a >' a . In other words, the tax schedule has 
to discourage investment enough so that individuals do not overinvest and violate the 



36 



incentive compatibility constraint of the ruler. Summarizing: 

Proposition 6 When citizens choose their investment levels individually, and the first- 
best level of investment cannot be supported, equilibrium taxes have to be distortionary 
to discourage citizens from investing up to e = 1 . 

Therefore, this simple model not only helps in analyzing the commitment problems, 
and the limitations of the PCT, but also suggests a reason for apparently inefficient 
methods of taxation (even when non-distortionary lump-sum taxes are available). These 
non-lump-sum taxes, at face value, appear to distort incentives. Nevertheless, once we 
are in the realm of self-enforcing agreements between rulers and citizens, an important 
consideration is to prevent citizens from overinvesting. Thus, the form of taxation has to 
be such that citizens do not have an incentive to invest too much, that is, citizens should 
not be the full residual claimant of the returns from their investments. At this stage, this 
explanation for why distortionary taxes may be preferred to non-distortionary alternatives 
is simply an implication of the model, and it is not currently clear how important this 
rationale is in practice. I leave a more detailed investigation of these issues to future work. 

6 Concluding Remarks 

There is growing interest and work on the determinants of policies and on the institutional 
choices that societies make. Why do some societies choose high taxes, while others opt 
for lower taxation? Why do some societies close their borders to trade, while others are 
more open? Why are bureaucracies more corrupt in some countries than in others? Why 
are some societies democratic, some parliamentarian, some majoritarian, some relying on 
common law, etc.? 

Much recent work is attempting to answer these questions. The first step towards an 
answer is to decide who makes the policy and institutional choices, and for whose interests. 
Or in other words, do collective choices maximize the output, surplus or welfare of the 
society as a whole, or do they select policies and institutions that benefit certain politically 
powerful groups, while being detrimental for aggregate output, surplus or welfare? Even 
though this latter question is, in some sense, antecedent to the ones posed above, there 
is not yet general agreement on the answers. 

This paper provides a simple taxonomy for the possible answers to this question. Either 
we could subscribe to what I dubbed the "Political Coase Theorem," which argues that 

37 



societies make efficient (output- or surplus-maximizing) choices, and distribute the gains 
from these choices between various groups and individuals. According to this approach, 
when societies choose inefficient policies, there will be strong political and social forces 
pushing them back towards efficient policies. Alternatively, societies may choose inefficient 
policies, not because of a failure in the political process, but because politicians' and 
citizens' beliefs were "mistaken" . Finally, we can be in the realm of Theories of Social 
Conflict, which maintain that societies often choose the wrong policies and institutions, 
or even pursue disastrous courses of action, because these choices are not made for the 
benefit of the society as a whole, but for the benefit of those who control political power. 

The bulk of the paper was devoted to arguing that Theories of Social Conflict provide 
the right empirical and theoretical framework for the analysis of the questions posed above. 
But why do politically powerful groups choose policies that reduce aggregate output 
rather than choosing the "right" policies, and redistributing the gains to themselves? The 
answer this paper developed is that there are serious commitment problems in politics 
placing severe limits on the reasoning of the Political Coase Theorem. In other words, 
efficiency considerations cannot be separated from distributional conflicts. The reasoning 
of the Political Coase Theorem presumes the possibility of political and economic trades 
between various individuals and groups. But these trades are intertemporal, and need to 
rely on contracts and promises. Typically, contracts and explicit promises are enforced by 
"the state". When it comes to contracts that the state, or social groups controlling the 
state, would like to write with the rest of the society, they will be non-enforceable. This 
implies that the allocation of political power creates an inherent commitment problem, 
imdermining the potential to reach efficient outcomes (naturally, this does not deny the 
fact that political and economic forces will sometimes push towards more efficient social 
arrangements). I investigated how incentive-compatible promises can make up for lack 
of enforceable contracts, but generally fall short of achieving the efficient outcome (or of 
validating the Political Coase Theorem). I also developed some comparative statics from 
the simple model I used for this purpose. 

This paper is only a preliminary attempt to highlight some of the important issues 
that are implicit in much of the recent political economy literature. I believe that the 
evidence is clear that Theories of Social Conflict provide the right framework for further 
analysis. But there are certainly factors other than commitment problems important in 
preventing the Political Coase Theorem from applying, and even if commitment problems 



38 



are what matter most, the way this paper modeled them may not be the most fruitful 
approach. So this paper may be thought of as a call for future research on investigating 
the major factors that cause inefficient policies and prevent the Political Coase Theorem 
from applying. 



39 



7 References 

Acemoglu, Daron, Philippe Aghion and Fabrizio Zilibotti (2002) "Distance to 
Frontier, Selection and Economic Growth" NBER Working Paper, No. 9066. 

Acemoglu, Daron, Simon Johnson and James A. Robinson (2001) "The 
Colonial Origins of Comparative Development: An Empirical Investigation," American 
Economic Review, volume 91, No. 5, 1369-1401. 

Acemoglu, Daron and James A. Robinson (2000a) "Why did the West Extend 
the Franchise? Democracy, Inequality and Growth in Historical Perspective," Quarterly 
Journal of Economics, CXV, 1167-1199. 

Acemoglu, Daron and James A. Robinson (2000b) "Political Losers as a 
Barrier to Economic Development," American Economic Review Papers and Proceedings, 
90, 126-130. 

Acemoglu, Daron and James A. Robinson (2001) "Inefficient Redistribution," 
American Political Science Review, v95, n3 (September 2001): 649-61. 

Acemoglu, Daron and James Robinson (2002) "Economic Backwardness in 
Political Perspective" NBER Working Paper, No. 8831. 

Acemoglu, Daron and Thierry Verdier (1998). "Property Rights, Corruption 
and the Allocation of Talent: A General Equilibrium Approach." Economic Journal, 
108, 1381-1403. 

Aghion, Philippe and Patrick Bolton (1990) "Government and the Risk of De- 
fault: a Political-Economic Model of Strategic Role of Debt" In Rudi Dornbusch and 
Mario Draghi, editors, Public Debt Management: Theory and History, Cambridge, Cam- 
bridge University press, 1990, 121-45. 

Anderlini, Luca and Leonardo Felli (1998) "Costly Coasian Contracts" London 
School of Economics working paper. 

Aldrich, John H. (1983) "A Downsian Spatial Model with Party Activism." Amer- 
ican Political Science Review 77: 974-90. 

Austen-Smith, David, and Jeffrey S. Banks (1999) Positive Political Theory I: 
Collective Preference. Ann Arbor MI: University of Michigan Press. 

Baron, David P., and John Ferejohn (1989) "Bargaining in Legislatures." Amer- 
ican Political Science Review 83: 1181-206. 

Becker, Gary S. (1983) "A Theory of Competition among Pressure Groups for 
Political Influence," Quarterly Journal of Economics, 98, 371-400. 

40 



Becker, Gary S. (1985) "Public Policies, Pressure Groups, and Dead Weight Costs," 
Journal of Public Economics, 28, 329-347. 

Becker, Gary S. and Casey Mulligan (1998) "Deadweight Costs and the Size of 
Government" NBER Working Paper No.6789. 

Besley, Timothy, and Stephen Coate (1998) "Sources of Inefficiency in a Rep- 
resentative Democracy: a Dynamic Analysis" American Economic Review, 88 139-156. 

Buchanan, James M., and Gordon Tullock (1962) The Calculus of Consent. 
Ann Arbor MI: University of Michigan Press. 

Coase, Ronald (1960) "The Problem of Social Cost" Journal of Law and Economics, 
vol. 3, pp. 1-31. 

Coate, Stephen and Stephen Morris (1995) "On the Form of Transfers to Special 
Interests," Journal of Political Economy, 103, 1210-1235. 

Coate, Stephen and Stephen Morris (1998) "Policy Persistence," American 
Economic Review. 

Cain, Philip J. and Anthony G. Hopkins (1993) British Imperialism.: Innovation 
and Expansion 1688-1914, Longman, New York. 

Coatsworth, John H. (1999) "Economic and Institutional Trajectories in 19th 
Century Latin America," in John H. Coatsworth and Alan M. Taylor eds. Latin America 
and the World Economy since 1800, Harvard University Press, Cambridge MA. 

Crowder, Michael (1968) West Africa under Colonial Rule, Northwestern Univer- 
sity Press, Evanston IL. 

Curtin, Philip D. (1964) Th,e Image of Africa, University of Wisconsin Press, 
Madison. 

Davis, Lance E. and Robert A. Huttenback (1987) Mammon and the Pursuit of 
Empire: The Political Economy of British Imperialism, 1860-1912, Cambridge University 
Press, Cambridge UK. 

Denoon, Donald (1983) Settler Capitalism: The Dynamics of Dependent Develop- 
ment in the Southern Hemisphere, Clarendon Press, Oxford UK. 

Demsetz, Harold (1967) "Toward a Theory of Property Rights" American Eco- 
nomic Review, volume 57. 

Dixit, Avinash, Gene M. Grossman, and Faruk Gul (2000) "The Dynamics 
of Political Compromise." Journal of Political Economy 108: 531-568. 

Dixit, Avinash arid John Londregan (1995) "Redistributive Politics and Eco- 



41 



nomic Efficiency," American Political Science Review, 89, 856-866. 

Dixit, Avinash and Mancur Olson (2000) "Does Voluntary Participation Under- 
mine the Coase Theorem?" Journal of Public Economics vol 76, n3, 309-35. 

Djankov, Simeon, Rafael, La Porta, Florencio Lopez-de-Silanes, and Andrei 
Shleifer (2002), "The Regulation of Entry," Quarterly Journal of Economics, CXVII, 

Evans, Peter, (1998) Embedded Autonomy: States and Industrial Transformation, 
Princeton University Press. 

Farrell, Joseph (1987) "Information and the Coase Theorem" Journal of Economic 
Perspectives vol 1, 113-29. 

Ferejohn, John (1986) "Incumbent Performance and Electoral Control," Public 
Choice 50: 5-25. 

Gann, Lewis. H. and Peter Duignan (1962) White Settlers in Tropical Africa, 
Penguin, Baltimore MD. 

Gann, Lewis. H. and Peter Duignan (1979) The Rulers of Belgian Africa, 
Princeton University Press, Princeton NJ. 

Glaesar, Edward and Andrei Shleifer (2002) "Legal Origins" forthcoming Quar- 
terly Journal Economics. 

Grossman, Gene and Elhanan Helpman (1994) "Protection for Sale," American 
Economic Review, 84, 833-850. 

Grossman, Gene and Elhanan Helpman, (2001), Special Interest Politics, MIT 
Press, Cambridge MA and London UK 

Grossman, Herschel and M-S. Kim (1995) "Swords or Ploughshares? A Theory 
of the Security of Claims to Property," Journal of Political Economy, 103, 1275-1288. 

Grossman, Sanford and Oliver Hart (1986) "The Costs and Benefits of Own- 
ership: a Theory of Vertical and Lateral Integration" Journal of Political Economy, vol. 
94, pp. 691-719. 

Ha-Cheong Yeon (1988) "Economy" in A Comparative Study of South and North 
Korea, edited by the national Unification Board, Seoul, Korea. 

Hall, Peter A. and David Soskice (2001) Varieties of capitalism : the institutional 
foundations of comparative advantage, New York; Oxford University Press. 

Hart, Oliver and John Moore (1990) "Property Rights and the Nature of the 
Firm" Journal of Political Economy Vol 98, pp 1119-1158. 

Herbst, Jeffery I. (2000) States and Power in Africa: Comparative Lessons in 



42 



Authority and Control. Princeton University Press, Princeton NJ. 

Knack, Steven and Philip Keefer (1995) "Institutions and Economic Perfor- 
mance: Cross-Country Tests Using Alternative Measures," Economics and Politics, 7, 
207-227. 

Krueger, Alan, and Mikael Lindahl (2001), "Education for Growth in Sweden 
and the World," Journal of Economic Literature. 

Kupperman, Karen Ordahl (1988) "England's Dream of Caribbean Empire" 
William and Mary Quarterly, vol 45, pp. 70-99, reprinted in Karen Ordahl Kupperman 
(editor) Major problems in American colonial history : documents and essays, Boston, 
MA : Houghton Mifflin, 2000. 

Libecap, Gary D. (1989) Contracting for property rights, Cambridge, Cambridge 
University Press. 

Maddison, Angus (2001) The World Economy: A Millenial Perspective, Develop- 
ment Centre of the Organization for Economic Cooperation and Development, OECD, 
Paris. 

Manning, Patrick (1982) Slavery, Colonialism, and Economic Growth in Dahomey, 
1640-1980, Cambridge University Press, New York NY. 

Mauro, Paulo (1995) "Corruption and Growth," Quarterly Journal of Economics, 
CX, 681-712. 

McKelvey, Richard D. and Talbot Page (1999) "Taking the Coase Theorem 
Seriously" Economics and Philosophy vol 15, n2, 235-47. 

Meltzer, Allan H. and Scott Richards (1981) "A Rational Theory of the Size of 
Government" Journal of Political Economy volume 89, #5, 914-927. 

Mukand Sharun and Dani Rodrik (2002) "In Search of Holy Grail: Policy Con- 
vergence, Experimentation and Economic Growth" Harvard Kennedy School mimeo. 

Myerson, Roger B (1995) "Analysis of Democratic Institutions: Structure, Con- 
duct and Performance" Journal of Economic Perspectives 9: 77-89. 

North, Douglass C. (1981) Structure and Change in Economic History, W.W. 
Norton k Co., New York. 

North, Douglass C. (1990) "A Transaction Cost Theory of Politics" Journal of 
Theoretical Politics 2(4). 

Olson, Mancur, (1965) The Logic of Collective Action, Harvard University Press, 
Cambridge MA. 



43 



Olson, Mancur, (2000) Power and Prosperity : Outgrowing Communist and Capi- 
talist Dictatorships, New York, NY, Basic Books. 

Peemans, Jean-Philippe (1975) "Capital Accumulation in the Congo under Colo- 
nialism: The Role of the State," in Lewis. H. Gann and Peter Duignan eds. Colonialism 
in Africa 1870-1960, Volume 4 The Economics of Colonialism, Hoover Institution Press, 
Stanford CA. 

Persson, Torsten and Guide- Tabellini (2000) Political Economics: Explaining 
Economic Policy, The MIT Press, Cambridge MA 

Persson, Torsten and Lars Svensson (1989) "Why a Stubborn Conservative 
Would Run a Deficit: Policy with Time-Inconsistent Preferences" Quarterly Journal of 
Economics, 104, 325-46. 

Piketty, Thomas, (1995) "Social Mobility and Redistributive Politics, Quarterly 
Journal of Economics, 110. 

Rajan, Raghuram G. and Zingales, Luigi (2000) "The Tyranny of Inequality" 
Journal of Public Economics vol 76, n3 521-58. 

Reno, William (1995) Corruption and State Politics in Sierra Leone, Cambridge 
University Press, New York NY. 

Ringer, F. (1979) Education and Society in Modern Europe, Indiana University 
Press, Bloomington. 

Roberts, Andrew (1976) A History of Zambia, Heinemann, London. 

Robinson, Ronald E. and John Gallagher (1961) Africa and the Victorians: 
The Official Mind of Imperialism, MacMillan, London. 

Robinson James A. (1998) "Theories of "Bad" Policy", (1998) Journal of Policy 
Reform, 3, 1-46. 

Rodrik, Dani (1986) "Tariffs, Subsidies and Welfare with Endogenous Policy," 
Journal of International Economics,", 21, 285-299. 

Romer, David (1997) "Misconceptions and Political Outcomes" NBER Working 
Paper No.6117. 

Romer, Thomas (1975) "Individual Welfare, Majority Voting and the Properties 
of a Linear Income Tax" Journal of Public Economics, 4, 163-85. 

Shleifer, Andrei and Robert W. Vishny (1998) The Grabbing Hand: Govern- 
ment Pathologies and their Cures, Harvard University Press, Cambridge MA. 

Snyder, James M., Jr. (1990) "Resource Allocation in Multiparty Elections." 



44 



American Journal of Political Science 34: 59-73. 

Spiller, Pablo and Mariano Tommasi (2002) "The Institutional Foundations of 
Public Policy: A Transactions Approach with Application to Argentina" Berkeley inimeo. 

Staiger, Robert W. and Guido Tabellini (1987) "Discretionary Trade Policy 
and Excessive Protection," American Economic Review, 77, 340-348. 

Stigler, George (1966) The Theory of Price, Third Edition, Macmillan. 

Tilly, Charles (1990) Coercion, Capital and European States, AD 990-1990, Black- 
well, Cambridge MA. 

Tullock, Gordon (1983) The Economics of Income Redistribution, Kluwer-Nijhoff, 
Boston MA. 

Wade, Robert (1990) Governing the Market: Economic Theory and the Role of 
Government in East Asian Industrialization, Princeton, Princeton University Press. 

Weingast, Barry R. (1998) "Political Stability and Civil War: Institutions, Com- 
mitment and American Democracy," in R.H. Bates et al. eds. Analytic Narratives, Prince- 
ton University Press, Princeton NJ. 

Wittman, Donald (1989) "Why Democracies Produce Efficient Results," Journal 
of Political Economy, 97, 1395-1426. 

Wittman, Donald (1995) The Myth of Democratic Failure: Why Political Institu- 
tions are Efficient, University of Chicago Press, Chicago IL. 

Wilson, James D. (1990) "Are Efficiency Improvements in Government Transfer 
Policies Self-Defeating in Political Equilibrium," Economics and Politics, 2, 241-258. 

Young, Crawford (1994) The African Colonial State in Comparative Perspective, 
Yale University Press, New Haven CT. 



45 



CM 



^5 

a © 



O 

£ 

u 



© 

s- 



a 



© s 

WD .© 
© +3 

• • »pN 

WD 2 







0_ 


< 

C/3 

3 


Z 
< 

o 


CO 




N 

2 


CO 
< 




oo 



1" 
oo 



f 

CD 



CD 



or 
< 



Q 
CO 



CD 
t 
O 



CD 

CU 
CO 

o 



00 



CM 



96-9861 >)S!cJ uojjeudojdx^ e6ej9AV 



<P TO 


0> CD 

9 o 


* £ 


3 -c 


o o 


co -z. 


\ 



5- o ^ 



o 



CM 



2 tt C 

a o § 

Q "D T3 

C -D 

.. H5 05 

CM JZ S 

£ 3 £ 

3 O 

O)C0 £ 

iZ c 




u I 
(33 



O 

IT, 



o 

CO 

cr> 



o 
r- 



o 
CD 



o 
m 
en 



o 


o 


O 


o 


o 


o 


o 


o 


o 


O 


o 


o 


o 


o 


o 


o 


o 


o 


o 


o 


o 


1- 


r J 


O 


CO 


CD 


-f 


CM 



C3 



cr 

• • 

WD 




i 



* _ 




* ^ 




« 






Ofl 




(GO 








a &d 




OD u 


o 


£3 ^ 




PQ 


* 




<U 





.55 .3 a 



C/3 



oo 



^ 



Date Due 



Lib-26-67 



MIT LIBRARIES 



I 



3 9080 02527 9750 



m